LucasBoTang/GradNorm

PyTorch implementation of the GradNorm

/ 100

Emerging

When training deep learning models that handle multiple related tasks simultaneously, it's common for some tasks to learn faster or slower than others, leading to an imbalanced model. This tool helps machine learning engineers and researchers automatically adjust the training process so that all tasks learn at a more balanced rate. You input your multi-task neural network, its loss function, and training data, and it outputs optimized task weights and training loss logs, improving overall model performance.

119 stars. No commits in the last 6 months.

Use this if you are developing or training deep learning models that perform multiple tasks at once and are struggling with balancing the training progress across these different tasks.

Not ideal if you are working with single-task deep learning models or do not require dynamic adjustment of task losses during training.

deep-learning-training multi-task-learning neural-network-optimization machine-learning-engineering

Stale 6m No Package No Dependents

Maintenance 0 / 25

Adoption 10 / 25

Maturity 16 / 25

Community 9 / 25

How are scores calculated?

Stars

119

Forks

Language

Jupyter Notebook

License

Apache-2.0

Higher-rated alternatives

fangwei123456/spikingjelly

SpikingJelly is an open-source deep learning framework for Spiking Neural Network (SNN) based on PyTorch.

neuromorphs/NIR

Neuromorphic Intermediate Representation reference implementation

BindsNET/bindsnet

Simulation of spiking neural networks (SNNs) using PyTorch.

norse/norse

Deep learning with spiking neural networks (SNNs) in PyTorch.

jeshraghian/snntorch

Deep and online learning with spiking neural networks in Python

Explore ML Frameworks

All categories Trending ML Framework directory Insights